Files
manual_slop/tests/conftest.py
Ed_ 35480a26dc test(audit): fix critical test suite deadlocks and write exhaustive architectural report
- Fix 'Triple Bingo' history synchronization explosion during streaming

- Implement stateless event buffering in ApiHookClient to prevent dropped events

- Ensure 'tool_execution' events emit consistently across all LLM providers

- Add hard timeouts to all background thread wait() conditions

- Add thorough teardown cleanup to conftest.py's reset_ai_client fixture

- Write highly detailed report_gemini.md exposing asyncio lifecycle flaws
2026-03-05 01:42:47 -05:00

321 lines
11 KiB
Python

import pytest
import asyncio
import subprocess
import time
import requests
import os
import signal
import sys
import datetime
import shutil
from pathlib import Path
from typing import Generator, Any
from unittest.mock import patch
# Ensure project root and src/ are in path for imports
sys.path.append(os.path.abspath(os.path.join(os.path.dirname(__file__), "..")))
sys.path.append(os.path.abspath(os.path.join(os.path.dirname(__file__), "..", "src")))
# Import the App class after patching if necessary, but here we just need the type hint
from gui_2 import App
class VerificationLogger:
def __init__(self, test_name: str, script_name: str) -> None:
self.test_name = test_name
self.script_name = script_name
self.entries = []
self.start_time = time.time()
# Route artifacts to tests/logs/
self.logs_dir = Path(f"tests/logs/{datetime.datetime.now().strftime('%Y%m%d_%H%M%S')}")
self.logs_dir.mkdir(parents=True, exist_ok=True)
def log_state(self, field: str, before: Any, after: Any) -> None:
delta = ""
if isinstance(before, (int, float)) and isinstance(after, (int, float)):
diff = after - before
delta = f"{'+' if diff > 0 else ''}{diff}"
self.entries.append({
"Field": field,
"Before": str(before),
"After": str(after),
"Delta": delta
})
def finalize(self, title: str, status: str, result_msg: str) -> None:
round(time.time() - self.start_time, 2)
log_file = self.logs_dir / f"{self.script_name}.txt"
with open(log_file, "w", encoding="utf-8") as f:
f.write(f"[ Test: {self.test_name} ]\n")
f.write(f"({title})\n\n")
f.write(f"{self.test_name}: before vs after\n")
f.write(f"{'Field':<25} {'Before':<20} {'After':<20} {'Delta':<15}\n")
f.write("-" * 80 + "\n")
for e in self.entries:
f.write(f"{e['Field']:<25} {e['Before']:<20} {e['After']:<20} {e['Delta']:<15}\n")
f.write("-" * 80 + "\n")
f.write(f"{status} {self.test_name} ({result_msg})\n\n")
print(f"[FINAL] {self.test_name}: {status} - {result_msg}")
@pytest.fixture(autouse=True)
def reset_ai_client() -> Generator[None, None, None]:
"""
Autouse fixture that resets the ai_client global state before each test.
This is critical for preventing state pollution between tests.
"""
import ai_client
import mcp_client
ai_client.reset_session()
# Reset callbacks to None or default to ensure no carry-over
ai_client.confirm_and_run_callback = None
ai_client.comms_log_callback = None
ai_client.tool_log_callback = None
# Clear all event listeners
ai_client.events.clear()
# Reset provider/model to defaults
ai_client.set_provider("gemini", "gemini-2.5-flash-lite")
# Reset MCP client state
mcp_client.configure([], [])
yield
ai_client.reset_session()
@pytest.fixture
def vlogger(request) -> VerificationLogger:
"""Fixture to provide a VerificationLogger instance to a test."""
test_name = request.node.name
script_name = Path(request.node.fspath).stem
return VerificationLogger(test_name, script_name)
def kill_process_tree(pid: int | None) -> None:
"""Robustly kills a process and all its children."""
if pid is None:
return
try:
print(f"[Fixture] Attempting to kill process tree for PID {pid}...")
if os.name == 'nt':
# /F is force, /T is tree (includes children)
subprocess.run(["taskkill", "/F", "/T", "/PID", str(pid)],
stdout=subprocess.DEVNULL,
stderr=subprocess.DEVNULL,
check=False)
else:
# On Unix, kill the process group
os.killpg(os.getpgid(pid), signal.SIGKILL)
print(f"[Fixture] Process tree {pid} killed.")
except Exception as e:
print(f"[Fixture] Error killing process tree {pid}: {e}")
@pytest.fixture
def mock_app() -> Generator[App, None, None]:
"""
Mock version of the App for simple unit tests that don't need a loop.
"""
with (
patch('src.models.load_config', return_value={
'ai': {'provider': 'gemini', 'model': 'gemini-2.5-flash-lite'},
'projects': {'paths': [], 'active': ''},
'gui': {'show_windows': {}}
}),
patch('gui_2.save_config'),
patch('gui_2.project_manager'),
patch('gui_2.session_logger'),
patch('gui_2.immapp.run'),
patch('src.app_controller.AppController._load_active_project'),
patch('src.app_controller.AppController._fetch_models'),
patch.object(App, '_load_fonts'),
patch.object(App, '_post_init'),
patch('src.app_controller.AppController._prune_old_logs'),
patch('src.app_controller.AppController.start_services'),
patch('src.app_controller.AppController._init_ai_and_hooks'),
patch('gui_2.PerformanceMonitor')
):
app = App()
yield app
if hasattr(app, 'controller'):
app.controller.shutdown()
elif hasattr(app, 'shutdown'):
app.shutdown()
@pytest.fixture
def app_instance() -> Generator[App, None, None]:
"""
Centralized App instance with all external side effects mocked.
Matches the pattern used in test_token_viz.py and test_gui_phase4.py.
"""
with (
patch('src.models.load_config', return_value={
'ai': {'provider': 'gemini', 'model': 'gemini-2.5-flash-lite'},
'projects': {'paths': [], 'active': ''},
'gui': {'show_windows': {}}
}),
patch('gui_2.save_config'),
patch('gui_2.project_manager'),
patch('gui_2.session_logger'),
patch('gui_2.immapp.run'),
patch('src.app_controller.AppController._load_active_project'),
patch('src.app_controller.AppController._fetch_models'),
patch.object(App, '_load_fonts'),
patch.object(App, '_post_init'),
patch('src.app_controller.AppController._prune_old_logs'),
patch('src.app_controller.AppController.start_services'),
patch('src.app_controller.AppController._init_ai_and_hooks'),
patch('gui_2.PerformanceMonitor')
):
app = App()
yield app
# Cleanup: Ensure background threads and asyncio loop are stopped
if hasattr(app, 'controller'):
app.controller.shutdown()
if hasattr(app, 'shutdown'):
app.shutdown()
# Use controller._loop for cleanup
loop = getattr(app.controller, '_loop', None) if hasattr(app, 'controller') else None
if loop and not loop.is_closed():
tasks = [t for t in asyncio.all_tasks(loop) if not t.done()]
if tasks:
# Cancel tasks so they can be gathered
for task in tasks:
task.cancel()
# We can't really run the loop if it's already stopping or thread is dead,
# but we try to be clean.
try:
if loop.is_running():
loop.call_soon_threadsafe(loop.stop)
except: pass
# Finally close the loop if we can
try:
if not loop.is_running():
loop.close()
except: pass
@pytest.fixture(scope="session")
def live_gui() -> Generator[tuple[subprocess.Popen, str], None, None]:
"""
Session-scoped fixture that starts sloppy.py with --enable-test-hooks.
Includes high-signal environment telemetry and workspace isolation.
"""
gui_script = os.path.abspath("sloppy.py")
diag = VerificationLogger("live_gui_startup", "live_gui_diag")
diag.log_state("GUI Script", "N/A", "gui_2.py")
# 1. Create a isolated workspace for the live GUI
temp_workspace = Path("tests/artifacts/live_gui_workspace")
if temp_workspace.exists():
for _ in range(5):
try:
shutil.rmtree(temp_workspace)
break
except PermissionError:
time.sleep(0.5)
# Create the workspace directory before writing files
temp_workspace.mkdir(parents=True, exist_ok=True)
# Create minimal project files to avoid cluttering root
# NOTE: Do NOT create config.toml here - we use SLOP_CONFIG env var
# to point to the actual project root config.toml
(temp_workspace / "manual_slop.toml").write_text("[project]\nname = 'TestProject'\n", encoding="utf-8")
(temp_workspace / "conductor" / "tracks").mkdir(parents=True, exist_ok=True)
# Resolve absolute paths for shared resources
project_root = Path(os.path.abspath(os.path.join(os.path.dirname(__file__), "..")))
config_file = project_root / "config.toml"
cred_file = project_root / "credentials.toml"
mcp_file = project_root / "mcp_env.toml"
# Preserve GUI layout for tests
layout_file = Path("manualslop_layout.ini")
if layout_file.exists():
shutil.copy2(layout_file, temp_workspace / layout_file.name)
# Check if already running (shouldn't be)
try:
resp = requests.get("http://127.0.0.1:8999/status", timeout=0.5)
if resp.status_code == 200:
print("[Fixture] WARNING: Hook Server already up on port 8999. Test state might be polluted.")
# Optionally try to reset it
try: requests.post("http://127.0.0.1:8999/api/gui", json={"action": "click", "item": "btn_reset"}, timeout=1)
except: pass
except: pass
print(f"\n[Fixture] Starting {gui_script} --enable-test-hooks in {temp_workspace}...")
os.makedirs("logs", exist_ok=True)
log_file_name = Path(gui_script).name.replace('.', '_')
log_file = open(f"logs/{log_file_name}_test.log", "w", encoding="utf-8")
# Use environment variable to point to temp config if App supports it,
# or just run from that CWD.
env = os.environ.copy()
env["PYTHONPATH"] = str(project_root.absolute())
if config_file.exists():
env["SLOP_CONFIG"] = str(config_file.absolute())
if cred_file.exists():
env["SLOP_CREDENTIALS"] = str(cred_file.absolute())
if mcp_file.exists():
env["SLOP_MCP_ENV"] = str(mcp_file.absolute())
process = subprocess.Popen(
["uv", "run", "python", "-u", gui_script, "--enable-test-hooks"],
stdout=log_file,
stderr=log_file,
text=True,
cwd=str(temp_workspace.absolute()),
env=env,
creationflags=subprocess.CREATE_NEW_PROCESS_GROUP if os.name == 'nt' else 0
)
diag.log_state("GUI Process PID", "N/A", process.pid)
max_retries = 15
ready = False
print(f"[Fixture] Waiting up to {max_retries}s for Hook Server on port 8999...")
start_time = time.time()
while time.time() - start_time < max_retries:
try:
response = requests.get("http://127.0.0.1:8999/status", timeout=0.5)
if response.status_code == 200:
ready = True
print(f"[Fixture] GUI Hook Server for {gui_script} is ready after {round(time.time() - start_time, 2)}s.")
break
except (requests.exceptions.ConnectionError, requests.exceptions.Timeout):
if process.poll() is not None:
print(f"[Fixture] {gui_script} process died unexpectedly during startup.")
break
time.sleep(0.5)
diag.log_state("Startup Success", "N/A", str(ready))
diag.log_state("Startup Time", "N/A", f"{round(time.time() - start_time, 2)}s")
if not ready:
diag.finalize("Live GUI Startup Telemetry", "FAIL", "Hook server failed to respond.")
print(f"[Fixture] TIMEOUT/FAILURE: Hook server for {gui_script} failed to respond.")
kill_process_tree(process.pid)
pytest.fail(f"Failed to start {gui_script} with test hooks.")
diag.finalize("Live GUI Startup Telemetry", "PASS", "Hook server successfully initialized.")
try:
yield process, gui_script
finally:
print(f"\n[Fixture] Finally block triggered: Shutting down {gui_script}...")
# Reset the GUI state before shutting down
try:
from api_hook_client import ApiHookClient
client = ApiHookClient()
client.reset_session()
time.sleep(0.5)
except: pass
kill_process_tree(process.pid)
time.sleep(1.0)
log_file.close()
# Cleanup temp workspace with retry for Windows file locks
for _ in range(5):
try:
shutil.rmtree(temp_workspace)
break
except PermissionError:
time.sleep(0.5)
except:
break